Tuesday, November 21, 2006

Well, at least he gave me an "A" once

Truth be told, I was not a very good law student. First, I hate lawyers. That's always a strike against you in law school. Second, I hate legal precedent. I mean, why should I reference some idiot's clearly bad decision 20 years ago just because he or she happens to have gotten appointed to the Supreme Court?

Be that as it may, I did do well in a few classes. Cass Sunstein's was one of them, and I always wondered why given that I rarely agree with him on much. (Plus, there was that time I accidentally implied his life resembled a country/western song. Hey, how was I to know his dog had just died??)

However, I'm starting to figure it out. I may not agree with Sunstein on much, but apparently he agrees with me! In a recent article by Sunstein & Bo Cowgill in The New Republic ("Can predictions markets forecast elections? Good Bet") , Sunstein and Cowgill argue that, even though online prediction markets (such as Tradesports.com) failed to predict the Nov. 7 Senate elections (as I noted here), they are still better at making predictions than pretty much anything else. Sunstein and Cowgill also rightly find the lack of faith of Reason Magazine's Katherine Mangu-Ward, the DailyKos' Markos Moulitsas, and Atrios disturbing.

That said, as Sunstein and Cowgill allude, the really interesting question is where and why prediction markets will consistently provide wrong answers. (And to think they won't is to have too much faith.) Some markets will have too few bettors and too little information. Those are easy. But are their other types of events that will consistently stymie prediction markets, for less obvious reasons? If you can figure out what those events are, you can make a lot of money. But then, of course, the prediction markets would be working, wouldn't they?

No comments: